Title: SSB frequency translation.
Post by: DL2JML on 06. June 2021, 15:14:06
I am posting this here and on mcHF@group.io, hence the English language. Answers in German are welcome.
On all bands, there is a spike at -12kHz below the signal when using -12kHz frequency translation (or at -6kHz, when using -6kHz, etc…). The power of this spike is independent of the SSB signal level, it is present when there is no signal. Because SSB power varies a bit when talking, that means that spike is about -20dB below a normal conversation signal, which is a bit high.
This spike is not present when not using frequency translation or when using FreeDV (which does not appear to use frequency translation, does it?), so I suppose that spike can be fixed in firmware. As the spike is not always present, it cannot be related to pin diodes. |
Title: Re:SSB frequency translation.
Post by: DF8OE on 06. June 2021, 15:54:03
It is the local oscillator signal, which is fed by parasitic capacitance within the mixer ICs to the PA (or in RX via capacitive coupling of the RF preamp directly to the antenna jack).
It is possible to reduce this by feeding a dc voltage to one of the mixer inputs. But sadly it is working frequency dependent (not surprising at all: it is due to internal capacitive coupling and capacitors do have frequency dependent impedance...) So you cannot get rid of this behaviour in complete. It is "normal behaviour" of mcHF.
Except CW everything is working in translated mode, so this "spike" is present at digital modes, too. It is present in non-translating, too: there you can find it at the working frequency.
I am writing in English because of much more readers can find answers for the same questions if the language is English ::).
vy 73
Andreas |
Title: Re:SSB frequency translation.
Post by: DL2JML on 06. June 2021, 20:48:57
The spike is not present when using FreeDV. Or maybe it is present, but within the main signal, as is the case when not using frequency translation.
Wouldn't it be possible to get rid of the spike by using pre-distortion? |
Title: Re:SSB frequency translation.
Post by: DF8OE on 07. June 2021, 05:11:17
Pre-Distortion would need a full duplex RX/TX. You have to receive your own signal and compare it to the af input. Full duplex is not possible with mcHF because of multiple reasons (only one codec, too low horse power on MCU and more).
If you develop a complete new rf board of course you can get rid off it!
Years ago my first idea was the same: doing some "small hardware improvements" and get rid off all these spurious signals and distortions. But the deeper I dug the more complications popped up. It is not possible to remove all unwantd signals automatically. If you want to do this you have to develop big hardware modifications which detect the unwanted signals so that you can get data for correcting them. The detection hardware would be much more complicated as the complete rf board. I saw this not a big taks: it is impossible. So I decided to add only the active elements (e.g. DAC converters at the mixers to add voltages from configuration data. This way is possible! But you get many, many new configuration elements in menu which are interacting which each other. So it would not be a linear approaching but an improvement where you can get optimal suppression for one spurious signal and at the same setting you increase another unwanted signal. You would get a huge mass of configuration elements which you must walk through interactively many times to get the best result. At the same time somenone else added only a few new settings: I/Q for more than two bands. Of course that gets better result in I/Q-errors! But nearly nobody uses these configuration elements ::). I have seen hundreds of mcHF (and clones) in the meantime and I cannot remember only a single one where these additional configuration was used - they all are "OFF" (but I have seen many where the mandatory I/Q correction for the only two bands was set to "OFF", too). So I decided to leave any attempt for such "improvements by manual configuration": nobody would use them. And another (unanswered) question still would have been to be explored: are there other dependencies except frequency ones, e.g. temperature, output power or supply voltage?
Your observation that the spike is not present in freeDV only gives two answers:
1) Your equipment cannot detect it, but it is there. In freeDV you have big output signal without any microphone signal. So your equipment must work with big level differences in near frequency span.
2) The signal is output power dependent. If you apply modulation it gets smaller.
I am tending to 1). Please check this in SSB using a modulated signal and compare it so freeDV.
mcHF is not a state-of-the-art SDR transceiver which gives you a perfect RX and TX quality, where you can add transverters and PAs and always get results comparable to a Hilberling or DUC/DDC TRX. It is a "small, inexpensive homebrew SDR TRX" - and this task it nearly does perfectly.
vy 73
Andreas |
Title: Re:SSB frequency translation.
Post by: DL2JML on 07. June 2021, 06:18:45
Well... I calibrated I/Q in all bands because I expected a state-of-the-art SDR transceiver... Maybe it is time to revise my plans.
|
Title: Re:SSB frequency translation.
Post by: DF8OE on 07. June 2021, 06:26:15
Yes - if this was your imagination you should revise it.
mcHF is perfect for portabl use. We have used it many times on fielddays. You can get good results with low budget and small, standalone functionality without need for any PC. It has interesting features like built-in CAT and audio via USB, an integrated CW keyer, a small LCD and has reasonable RX sensitivity. That's what it wants to be - and what is definitly true.
Signal quality in TX is "acceptable" (more or less). If you use it in QRP I judge "it is acceptable". For using transverters and/orPAs it is not (my opinion). But that's not the task it was constructed for...
vy 73
Andreas |
Title: Re:SSB frequency translation.
Post by: satoryboy on 08. June 2021, 04:10:35
To DO1JML
There is an expensive option for a good mobile standalone SDR - the KX3. There is a budget one - https://disk.yandex.ru/d/KE4KSyBF41weLw |
Title: Re:SSB frequency translation.
Post by: DL2JML on 08. June 2021, 06:14:04
@satoryboy
I am not going to click on an unknown zip from Yandex. I am also not so sure the KX3 has much better signal purity than the mcHF, even commercial equipment has limitations.
@DF8OE
I mainly wanted to say that I ran the IQ calibration on all bands, so that such people do exist.
As to the mcHF: indeed it fits the bill and I am both very happy that I managed to assemble one and very thankful for the work of the developers.
But my main motivation for the mcHF is that it supports FreeDV. I would like to try that, and hoped I could also try it with the mcHF connected to an external PA when needed. |
Title: Re:SSB frequency translation.
Post by: DF8OE on 08. June 2021, 08:49:31
I am sure that there are people who have walked through complete IQ calibration. But they are less than 1% of all users (thats my impression) and with every new configuration entry the number decreases... You have now done the best to get a good IQ failure suppression!
I have used freeDV some years ago and it worked. Due to a lack of QSO partners I have not done any freeDV QSO in the last time...
Signal purity is ok on the lower bands (below 18MHz) if you replace the PIN-diode switching by a relay. Of course you do have the LO spike and intermixing products.
Above 18MHz intemixing and distortions increase signifivant so I never would use a PA there.
Mixing concept does have some physical issues which are system issues so you cannot get rid of them. And you are right: commercial products are not from Superman, too ;D.
vy 73
Andreas |
Title: Re:SSB frequency translation.
Post by: DL2JML on 08. June 2021, 20:36:48
I am coming back with my question, because I am still not perfectly convinced.
Your explanation is that the spike is due the local oscillator signal, which would be fed by parasitic capacitance. That is possible.
But this does not explain why I don't see the spike when using FreeDV, unless FreeDV does not use frequency translation. My equipment can detect the spike using the test tone signal or using a fixed SSB signal of high level, so it can detect "big level differences in near frequency span". Neither does the spike appear to be dependent on power level on SSB. |
Title: Re:SSB frequency translation.
Post by: DF8OE on 09. June 2021, 06:13:13
Sorry - I do not have any other explanation and no time to dig into this.
Technically it s impossible that the spike can be influenced by firmware (input I and Q). So there is no explanation why you cannot observe it in freeDV. It must be there. It may be that you observe a complete different spurious signal.
If you do not put a signal on microphone ideally I and Q both are silent. So the spike which I am talking about is fed in both translated and non translated to the PA. The only difference between translated and non tranlated is where you can find it regarding the frequency where you have tuned your device to. In non translated mode you will find it exactly at the working frequency and in non translated it is shifted by the translation.
FreeDV is produced in SSB (USB or LSB) what you can see at the screen by "DI-U" and "DI-L". The mode is in firmware handled in SSB and freeDV identically, except the signal source. In SSB it is I and Q which is calculated by MCU and in freeDV mic first goes through a "modem routine" using the freeDV procedure and the I and Q are calculated. Nothing touches the spike.
I think you first should check if we are talking about the same signal.
The spike is very low at low frequencies and increases if you increase frequency. You can observe it at the antenna jack in TX and in RX (!!).
I am no longer working (and using) mixing concept - I am developing OVI40 DUC/DDC PCB which does not have such problems (thats the reason I left mixing concept 2.5 years ago). So I cannot give you further hints without digging again deeper (what I don't want to do)...
vy 73
Andreas |
Title: Re:SSB frequency translation.
Post by: DL2JML on 09. June 2021, 09:40:03
@DF8OE
I perfectly understand if you say you have no time to dig into this. You do not have to respond to all questions. This is a public forum, I asked the question to the community and not to you in particular. Maybe others have other ideas.
As to the idea that the signal "must be there", I will redo the measurements, but to the best of my knowledge it is not there. Why, I do not know. If I knew, I would not have asked the question in the first place. |
Title: Re:SSB frequency translation.
Post by: satoryboy on 09. June 2021, 10:35:32
| An hour ago I made the adjustment of the next pre-production test copy of my TPX Amber project. In the current firmware version, the mirror channel suppression is adjusted on each band, including WARS (two alignment points for 10m) - the actual suppression of the carrier and the mirror channel is no worse than 50 dB on each band. There is one more trick - alternative power level control, which radically improves carrier suppression at powers less than 5 watts. I have already given a link to the documentation. Of course, it is infected with the worst computer viruses, fear. However, I myself dream of ICOM IC-7610, but I cannot afford this device yet. 73! |
Title: Re:SSB frequency translation.
Post by: DF8OE on 09. June 2021, 10:38:34
As you see we do have many, many registered members but only a handful are actively writing. This is normal behaviour in our recent times ;D...
One more hint:
The amplitude of the spike is not correspondent to the TX power. I think you should test on 10m because the spike is there very powerful.- switch TX power to 5W
- step to PA menu
- activate "reduce level on high bands"
- reduce 5W PWR adjust to minimum (3)
- exit PA menu
now switch TX power to 0.5W
Use this configuration to take a look at the spike. If it is the one I am talking about it should not have changed its amplitude. You can check in SSB because there is no "high level signal" in the near if you do not inject any audio.If you now transmit in freeDV the level difference is much lower and the spike can be detected easier.
vy 73
Andreas
|
Title: Re:SSB frequency translation.
Post by: satoryboy on 09. June 2021, 11:11:57
| I conducted a series of experiments, carrier suppression depends on the level of the low frequency signal, temperature, tuning frequency. But something can really be done. This problem can be radically solved by switching to the DDC / DUC architecture, but this is already the next project, TRX Axis. |
Title: Re:SSB frequency translation.
Post by: DL2JML on 09. June 2021, 11:12:45
@satoryboy
I did not say your zip if full of viruses, I said that am not going to click on an unknown zip. At least, not without some explanations as to what may be within.
The explanations I would have liked to have is that the zip contains a description of the amber project, a mcHF look-alike with (hopefully) better RF design. Or you could have pointed to your github account here: https://github.com/satoryboy
Anyway: sorry for the misunderstanding and thank you for the info about the amber project. I may come back to it if I am not satisfied by the mcHF. But for the moment, I will try to operate the TX I just built... ;) |
Title: Re:SSB frequency translation.
Post by: DL2JML on 09. June 2021, 11:17:37
@DF8OE
You posted:The amplitude of the spike is not correspondent to the TX power. I think you should test on 10m because the spike is there very powerful.
switch TX power to 5W
step to PA menu
activate "reduce level on high bands"
reduce 5W PWR adjust to minimum (3)
exit PA menu
now switch TX power to 0.5W
Use this configuration to take a look at the spike. |
|
If I understand correctly, you are suggesting that I redo the measurement by lowering the transmitted signal to the lowest possible level? |
Title: Re:SSB frequency translation.
Post by: DF8OE on 09. June 2021, 11:21:26
Yes. The reason is that the spike amplitude does not depend on output level. If you check in SSB it is not problematic because of your output signal is nearly ZERO if you TX without injecting AF. But in freeDV you do have maximum RF power with no AF signal applied! This makes measurement of the spike very difficult. Signals can intermix, intermodulate, overlap and so on.
vy 73
Andreas |
Title: Re:SSB frequency translation.
Post by: DF8OE on 09. June 2021, 15:28:45
@satoryboy:
Your GitHub is completely unusable. UHSDR is over 2 years old and amber project is empty. And in the zip-file you distribute I am missing the firmware sources or a link to them (for each version you publish a binary). This is related to bootloader, too. Please stop distributing this zip or add a text file with download links for the firmware or place all the sources in the zip file. GPL makes it mandatory to do so. The easiest way is to distribute via GitHub (thats why 99% of all GPL or compareable licensed software is distributed via GitHub or Sourceforge). Because of you have to publish the source code for all versions (included older ones) these two possibilities make life much easier.
You can edit your own post and delete / replace the zip with one that fits the GPL.
vy 73
Andreas |
Title: Re:SSB frequency translation.
Post by: satoryboy on 09. June 2021, 17:24:22
| Thank you for your comment. The volume of the source code is very large, I was sure that it would be enough to upload the source code with each new version. Of course, I will not delete anything, I will just add links to the source code to the archive. And I will update the list. I'll do it right now. Thanks again. |
Title: Re:SSB frequency translation.
Post by: DF8OE on 09. June 2021, 18:40:26
That would match the license Yuri. Thank you for understanding! The background is that if anyone picks up the zip he must easily get the sources to build them.
vy 73
Andreas |
Diskussions- und Newsboard des DARC-Ortsverbandes I40 | Powered by YaBB SE
© 2001-2003, YaBB SE Dev Team. All Rights Reserved.
|